It is well known that systems for reproducing images by means of binary image data processing usually differ in their recording densities (recording density is defined as the number of recording pixels consisting of solid and transparent dots in a certain area) respectively. Therefore an image reproducing system of a certain recording density must cope with input density data (corresponding to the number of input pixels consisting of solid and transparent dots) exceeding the capacity of that of the image reproducing system by converting the input density data to permissible recording density data.
In order to carry out such a density conversion process, there exists a method as shown in FIG. 9 by which every several pixels (marked by circles) of all the input pixels in the main and the sub-scanning directions (refer to FIG. 9(a)) are picked up to be used for recording a reproduction image as shown in FIG. 9(b). This assumes that all the pixels picked up in the input scanning process are utilized for recording. In the particular case of FIG. 9, four pixels (circled) out of twenty-five pixels (five pixels by five pixels) are picked up for the recording process.
However, this method is disadvantageous in reproducing a thin line like a rule establishing a line as shown in FIG. 9(b) intermittent or different from the original shown in FIG. 9(a). In order to resolve this drawback, Japanese Patent Laid Open No. 58-215165 provides a method by which a signal indicative of high and low levels for controlling a recording beam is obtained by comparing the average density value of a group of input pixels on each of which appropriate weight coefficients are imposed with a corresponding reference value.
However, the above-mentioned method, being capable of converting only data density in the sub-scanning direction, cannot be applied to main scanning. In addition, comparison between the average density and the reference density requires complex circuitry.